IP Address Characterizing Based On Netflow

With the rapid development of network technology,the scale of Internet in China is getting bigger and bigger,and the traffic behavior of IP address is becoming more and more complicated.How to observe and describe the traffic behavior of IP address has draw attention of various research institutions.At present,there are few researches on the behavior of IP address traffic,and the system of traffic behavior measurement is lacking.In this paper,the IP address traffic behavior model is established by referring to the model and spectrum research method in ecological behavior.The concept of IP address traffic behavior spectrum is proposed,and an IP address character mining method based on feature spectrum is proposed.In the establishment of the IP address traffic behavior model,the traffic behavior of each IP address is split into separate non-intersecting traffic activities.Each activity is described by measures from four attributes of time,space,category,and intensity.The IP address traffic behavior feature is defined as the abstract of the IP address traffic behavior,which is expressed as the specific value of the measure,and the method of extracting the feature on each measure is given by analyzing the form of the measure.Then,the IP address of a network is compared to a biological population,and the traffic behavior characteristic spectrum is defined as the full record of the stable traffic behavior feature of the IP address in this network..The size of the feature spectrum is controlled by numerical signatures and feature stability screening methods.In the IP address character mining based on the feature spectrum,the IP address character is defined as the set of all the behavior features of the address,and an IP address role discovery and updating algorithm applied to the network traffic record data flow is designed.Finally,the actual data is used to verify the algorithm,which proves that the method can find the IP address role with sufficient semantics.